Editorial – February 2023
Well, I promised you a February issue, so here it is. It is a bit thin on the actual books read, though I have included a history book that I read for one of the LGBTQ+ History Month talks I gave. But I have seen a couple of movies and I have some other things to talk about.
Social media over the past month has been abuzz with the issue of supposed-AI software such as ChatGPT. I say “supposed” because these programs are not actually intelligent. They are just good at pattern matching and replication. If you feed them enough web pages with the text, “2 + 2 = 5” they will assert that as fact, even if, at the same time, they assert that 5 – 2 = 3. If producing academic material with citiations, they may make them up because they don’t understand the difference between a real paper and an imaginary one.
However, these tools are now good enough at both image and language processing to pose significant threats to the livelihoods of both artists and writers. I had a few things to say about the art issue last year. As for the written word, goodness knows where we will end up. Here are a few bits of data.
You have probably all heard about how Clarkesworld had to close submissions because their slush readers were overwhelmed with stories “written” by ChatGPT. Such material is generally easy to spot, but sifting through it takes time and energy. Because generating new stories is so quick and easy, bad actors can overwhelm the submissions process of any magazine and prevent its staff from getting on with their work. There are solutions, but they will necessarily involve pre-vetting who is allowed to submit, which is far from ideal.
One of my friends reported on Facebook that a fairly lucrative gig she had writing content for a website vanished overnight because the site owner claimed that he could get all of his content from ChatGPT now. I understand that major outlets such as CNET and Buzzfeed are now using ChatGPT to generate content, though both admit that its output has to be checked for errors.
The Kindle market is full of rapidly written books claiming to teach you how to make your fortune using ChatGPT. I gather than TikTok has a similar infestation. I suspect also that a significant proportion of the self-published books on Kindle will soon be “authored” by ChatGPT.
Of course not everyone sees these tools as a threat. The latest Kobo newsletter pointed me at this blog post about how to use these tools in a more constructive and responsible way.
The big question is where all this will end up. Given that the major publishers, just like every other large corporation these days, see themselves as being in the business of delivering value for shareholders, I am sure that their senior management is salivating over the prospect of being able to do away with writers altogether and produce endless streams of books “in the style of” whichever dead successful writer they have the rights to publish. I wonder if Asimov ever considered that his career would be continued long after his death by robots trained to immitate his writing.